November 16, 2026 — today in AI.
1. Nvidia China Curbs Could Hit AMD and AI Chip Stocks
New U.S. AI chip export restrictions could complicate Nvidia, AMD, and the wider semiconductor industry's global growth. Teams planning GPU procurement or cloud capacity outside the U.S. should expect tighter availability and possible shifts in pricing for China-exposed supply chains.
2. As the 'AI Empire' Wobbles, Google, Amazon and Microsoft Accelerate In-House Chip Development
Cloud providers are using their own accelerators to narrow the gap with Nvidia and strengthen their bargaining power over Nvidia GPU pricing. For builders, this points to a growing set of cloud-specific chips where price-performance may diverge from Nvidia-based instances, so workload benchmarks need to be run per platform.
3. OpenAI Signs on for 457MW AI Capacity with TensorWave
OpenAI has contracted 457 MW of AI compute capacity from TensorWave, a neocloud provider, adding to its earlier 20 MW agreement with TecFusions for sites in Pennsylvania and Arizona. Large, long-term reservations like this tighten available GPU capacity for smaller buyers and show frontier labs diversifying beyond hyperscalers.
4. Wärtsilä 790 MW Texas Data Center: 42 Gas Engines
A Texas data center plan calls for 790 MW of off-grid capacity powered by 42 Wärtsilä 50SG natural gas engines. Off-grid gas generation is becoming a practical way around grid interconnection delays, which makes power availability, not just chip supply, a binding constraint on where inference capacity can be built.
5. Over 400 American AI Data Centers Use As Much Electricity As Wisconsin, Michigan, or Indiana
A Nature Sustainability study quantifies electricity use and carbon emissions across more than 400 U.S. hyperscale AI data centers, with combined consumption comparable to several entire states. Expect growing scrutiny of power sourcing and permitting, which feeds directly into the cost and timeline of new inference capacity.